initial script for automating the creation of a controlled testing en… #2057
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…vironment for OOVs
reuben
reviewed
Dec 29, 2021
create_oovs.sh
Outdated
| #!/bin/bash | ||
| set -e | ||
|
|
||
| stag=1 |
create_oovs.sh
Outdated
| exit 1 | ||
| fi | ||
|
|
||
| step=1 |
There was a problem hiding this comment.
I guess this shouldn't be hardcoded? Or is it meant as a development tool, so you iterate on the parts as you get them working?
create_oovs.sh
Outdated
| echo "Step 1: Preparing Data" | ||
| if [ $step -le 1 ]; then | ||
|
|
||
| # Extract corpus unique vocabularies |
There was a problem hiding this comment.
Suggested change
| # Extract corpus unique vocabularies | |
| # Extract corpus vocabulary (unique words) |
create_oovs.sh
Outdated
| sed 's/ /\n/g' tmp/data.txt | sort | uniq -c | sort -nr > tmp/vocab.txt | ||
| grep -o . tmp/vocab.txt | sort -u > tmp/alphabet.txt | ||
|
|
||
| # Pick the least frequent 10% vocabularies to represent OOVs |
There was a problem hiding this comment.
Suggested change
| # Pick the least frequent 10% vocabularies to represent OOVs | |
| # Pick the least frequent 10% words to build OOV set |
create_oovs.sh
Outdated
| grep -o . tmp/vocab.txt | sort -u > tmp/alphabet.txt | ||
|
|
||
| # Pick the least frequent 10% vocabularies to represent OOVs | ||
| oov_count=$(wc tmp/vocab.txt | awk '{print int($0*0.1)}') |
There was a problem hiding this comment.
use wc -l to communicate intent earlier.
Suggested change
| oov_count=$(wc tmp/vocab.txt | awk '{print int($0*0.1)}') | |
| oov_count=$(wc -l tmp/vocab.txt | awk '{print int($0*0.1)}') |
There was a problem hiding this comment.
Size of OOV set should be a parameter (with default).
create_oovs.sh
Outdated
|
|
||
| # Prepare OOV csv for testing purposes (to assess imporvements on it) | ||
| grep -wFf tmp/oov_sents tmp/data.txt > tmp/oov_corpus.txt | ||
| grep -wFf tmp/oov_sents $data | sed '1 i\wav_filename,wav_filesize,transcript' > tmp/oov_corpus.csv |
There was a problem hiding this comment.
This sed command doesn't work on macOS:
sed: 1: "1 i\wav_filename,wav_fi ...": extra characters after \ at the end of i command
Can we make it portable to BSD sed? This fix worked for me:
Suggested change
| grep -wFf tmp/oov_sents $data | sed '1 i\wav_filename,wav_filesize,transcript' > tmp/oov_corpus.csv | |
| echo "wav_filename,wav_filesize,transcript" > tmp/oov_corpus.csv | |
| grep -wFf tmp/oov_sents $data >> tmp/oov_corpus.csv |
create_oovs.sh
Outdated
| fi | ||
|
|
||
| # Generate LM | ||
| echo "Step 2: Generaing Language Model" |
There was a problem hiding this comment.
Suggested change
| echo "Step 2: Generaing Language Model" | |
| echo "Step 2: Generating Language Model" |
create_oovs.sh
Outdated
| gzip -c tmp/scorer_corpus.txt > tmp/scorer_corpus.txt.gz | ||
| grep -vf tmp/oov_sents $data > tmp/scorer_corpus.csv | ||
|
|
||
| # Prepare OOV csv for testing purposes (to assess imporvements on it) |
There was a problem hiding this comment.
Suggested change
| # Prepare OOV csv for testing purposes (to assess imporvements on it) | |
| # Prepare OOV CSV for testing purposes (to assess improvements on it) |
create_oovs.sh
Outdated
| echo "Evaluating on OOV testing set." | ||
| python -m coqui_stt_training.evaluate --test_files tmp/oov_corpus.csv \ | ||
| --test_output_file tmp/results/oov_results.json --scorer_path native_client/kenlm.scorer \ | ||
| --checkpoint_dir /home/aya/work/tmp/AM/coqui-stt-1.1.0-checkpoint --test_batch_size $nj |
There was a problem hiding this comment.
Checkpoint path should be made into a parameter.
create_oovs.sh
Outdated
| if [ $step -le 3 ]; then | ||
| echo "Evaluating on OOV testing set." | ||
| python -m coqui_stt_training.evaluate --test_files tmp/oov_corpus.csv \ | ||
| --test_output_file tmp/results/oov_results.json --scorer_path native_client/kenlm.scorer \ |
There was a problem hiding this comment.
The native_client/kenlm.scorer should be kenlm.scorer, according to the command in the step above, right? And that should probably be changed to tmp/kenlm.scorer to keep all the outputs of the script contained to that folder.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.